Ah Geez - the Hard Drive Does Make a Difference

[sandyk]

David

I even go to the trouble of switching off the front and rear PWM controlled fans during ripping with EAC.

Not only do the large pulses reflect right through the PSU, thay make quiet aftermarket fans sound noisier than when using a pure DC supply.

These large pulses may affect the utimate stability of the typical 14.318MHZ master oscillator despite additional voltage regulation ?

Nlote that I am not stating this as fact, just posing the question.

 

I get better results when using the +5V Linear PSU (photo attached,hopefully) to supply high quality +5V Linear power, instead of noisy internal +5V SMPS

to the Corsair Voyager where I rip .wav files to,

and also now DL high resolution .flac files to.I also play the decoded high resolution .flac files from there.

( I have 1 x 32GB Corsair Voyager, and 3 x 16GB Corsair Voyagers)

I have had considerable input from other people regarding the perceived improvements,both direct from my W7 PC,

and from overseas contacts via numerous comparison DLs, so I am not suffering from a placebo effect !

SandyK

 

N.B.

This is NOT a commercial product, but DIY.

 

 

 

 

 

 

 

If this link does not work, hopefully Chris will not have a problem with me attaching the direct link to a thread in another forum ?

Alex

 

http://rockgrotto.proboards.com/index.cgi?board=talk&action=display&thread=6068

 

 

 

[khollister]

So you are asserting that power characteristics for a drive where you write the digital file resulting from ripping a track affects the data? And I assert it has to be affecting the actual data since you claim these differences are audible even after the file is copied/transmitted across the internet onto different playback systems.

 

I am a pretty open minded guy on this kind of stuff, but I am really tempted to throw the BS flag on this (BTW, this is not in any way a personal judgment on Alex - I accept that he believes he has heard the effects he writes about). Have you compared checksums of the two versions of the file (with and without your PS)? If they are different, I might believe that the noise from the power to the drive is affecting the ripping process somehow, thus resulting in non bit-perfect rips, but the implied explanation that the drive characteristics alter the sound of a file copied to it is stretching credulity.

 

Of course this brings up the whole issue of audible differences in ripped files even though both appear to be bit perfect via secure rippers and checks with AccurateRip.

 

The problem I have with this is it invariably leads to a conclusion that computers and, therefore, our worldwide infrastructure doesn't work because 2 copies of the same data don't represent the same information.

 

Before we even get there, I have a problem postulating a plausible reason why changes in digital data (as in ripping an audio track from 2 different ROM drives) present themselves as minor audible differences (e.g. "less air"). Any errors in the data stream would have to be stochastic and thus have a far greater chance of resulting in changes to the data that are not correlated. This would mean you would have a far greater chance of hearing clicks, pops or bursts of noise than the types of things mentioned.

 

If the ripping process involved conversion back to an audio analog signal somewhere in the chain, I would be fine with this, but it doesn't.

 

I am really struggling with this (but willing to change my mind if someone can offer a theory, no matter how wild).

 

[RankStranger]

Magic?

 

RS

 

[losingmyreligion]

 

 

"Any sufficiently advanced technology is indistinguishable from magic"

 

 

 

[khollister]

" Magic?"

 

Exactly - and I'm not quite ready to go there. Not looking for proof, just a theory as to how this can happen. Among other things, I am stunned that data changes when copied to a disk drive depending on the power characteristics of that drive, but the sound of the resultant files (which implies the data integrity) survives transmission through the internet and copying onto someone's drive on the other end. This goes way beyond debates about why cables sound different, for instance.

 

I can only come to a conclusion that there must be some interaction of the power noise spectrum on the ripping process, not on the copying process. But if the rips checksum the same and pass AccurateRip with and without the power supply, I'm baffled again.

 

Never mind that the effects are described in a way that appears to be deterministic, not stochastic, which is also pretty unbelievable in the digital domain. In other words, I would have to believe that whatever errors are causing corruption of the digital stream are only affecting the LSB's of the 16/24 bit word, not the MSB's (which would create more drastic errors in the sound).

 

I just can't get my head around this. Would love someone like Barry or Gordon to weigh in here with their opinions.

 

[jhwalker]

There is no theory that explains this. It is possible for two *very* different files to have the same checksum (statistical theory allows for this), but not for two almost identical files to do so (if there is *any* difference, the checksum is different). So the files must be identical; therefore, they *cannot* sound different when played back on the same equipment.

 

A audible difference whereby one file has "more air" or whatever would imply a change of literally every bit in the file, and the files would neither be the same size nor would the checksums match. So any effect heard in comparing the two files is an artifact of the testing process, and cannot be real.

 

This is an area where there is no wiggle room. I certainly agree the overall electrical environment impacts the output, but not that identical source data can produce different outputs based on the origin of the data.

 

[khollister]

One slight correction - the type of changes we are talking about would require a change in the low order bits of every 16/24 bit word of the file. Changing every bit would probably result in noise

 

[bernardl]

Hi Dmccombs,

 

What dac were you using and what protocole was used between the mac mini and the dac?

 

Was that synchronous or asynchronous?

 

Regards,

Bernard

 

 

[dsnyder]

Hello SandyK,

 

Quote:

 

I even go to the trouble of switching off the front and rear PWM controlled fans during ripping with EAC. Not only do the large pulses reflect right through the PSU, thay make quiet aftermarket fans sound noisier than when using a pure DC supply. These large pulses may affect the utimate stability of the typical 14.318MHZ master oscillator despite additional voltage regulation? Note that I am not stating this as fact, just posing the question.

 

Hmmm, ripping is an entirely separate activity not related to playback. I don't even use the same computer for ripping as I do for playback. Do you?

 

Those fans are there for a reason and should be on during ripping to keep your computer and drives cool. If things get too hot, your computer may try to protect itself by slowing down the CPU. That's not good for a real-time activity like ripping and can even damage your system. If physical vibrations or electrical/EM noise from fans are causing your computer to produce inconsistent rips, you'll want to find out why and get that fixed before bothering to rip more CDs. If that's too hard, you can always ship your music collection off to a service to have them ripped for you. Most charge around $2 or so per disc plus s/h.

 

As far as playback goes, I think it's cool that you have discovered improved sonics by using an external PSU with your thumb drives. The difference in sound, however, almost certainly has nothing to do with the PSU's affect on the performance of the thumb drives; the O/S would have let you know if it was experiencing read errors. Rather, the most likely explanation for the difference in sonics has to do with the way that introducing the PSU has affected electrical noise levels inside your computer or the RF noise levels emitted by it, or both. If you can eliminate the noise source completely or isolate your PC from the downstream components that are affected by it, I think that you'll see even more improvement.

 

-- David

 

 

[sandyk]

Hi David

Give me a little credit! I am very mindful of temperature rise, and do not rip for extended periods,especially on hot days,when I avoid ripping at all ! Unlike many Macs, this Windows PC has more natural ventilation and doesn't need huge amounts of cooling.I also have fitted a Red LED to remind me to

turn the switch back on again.If ripping compilations, the fans are always turned on again for a while between changes of CD.

There are several members here who have already heard some of my comparison .wav files from over a year ago and posted about hearing the differences between .wav files with identical check sums. Although I took a break from posting here for over a year, I have been in regular contact with several of them, and uploaded more comparison .wav files , with one of the files using the USB pen and Linear PSU method. I even tested these comparisons with a NYC member during development of the Linear USB Power supply.

Earlier rips using just a USB Power Isolator as published by Silicon Chip magazine were reported back as sounding a little too HF prominent.

Did you read PeterSt's recent reply to me ?

Peter has already posted in CA some time back that he also heard those differences.

"Sat, 10/24/2009 - 05:39 — PeterSt Memory

Hi Sandy - All played from memory (and gapless hehe) :

 

.wav, .flac, .aif, .aiff, .mp3

 

16/22050*^

24/22050*

16/44100*^

24/44100*

16/48000*^

24/48000*

16/88200

24/88200+

16/96000

24/96000+

24/176400

24/192000

24/352800 (natively, or downsampled to 24/176400 incl. anti aliasing when the DAC can't cope)

 

All is output in 32 bits for DACs > 16 bits, unless the DAC accepts 24 bits only (rare but exists).

 

*) Also one channel, turned into two channel mono.

 

+) Can also output 16 bits. Thus, file is 24 bits, but playable on 16 bit DACs. I'm not sure whether 176400 and 192000 can do this too, so I left those out for safety.

 

^) Can output upsampled to double or quadrupule the input rate, in 24 (32) bits by means of Linear Interpolation (bad), sincx filtering (soso) or Arc Predition (good).

 

HTH !

Peter

 

PS: And since we're in touch here : Yes, your files sound different, and they are played from memory. I'm still recovering from this finding, but will setup the measuring environment now (at last). We were with two, and listening through "normal" loud speakers. Both heard the same differences. "Track04" is the more delicate one.

Heading for a job outside of IT now. ;-)"

 

 

How many of you have even tried ripping directly to a USB pen plugged directly into the PC's backplane. Just like the reports of improved SQ of SSD over HDD, the same applies, and for similar reasons.

Don't try it with a laptop though, unless you run on battery power ! Many people report hum and other noise from some laptops via USB, which is another reason to use an external PSU.

 

Alex

 

[dsnyder]

Howdy,

 

Gosh, I don't know what to say. I love you guys...

 

-- David

 

[ItemAudio]

It's tough to imagine how a systemic, qualitative change in an audio file could be generated by differences between rips. If they sum, they're the same. Errors would at best/worst be random and punctiliar. Eventually we come to some level where bits really are bits – so I'm currently rip-skeptic.

 

But. As with much else in digital audio, 'bits are bits' is only germane at a really esoteric - almost platonic - level. Bits depend on all kinds of mucky mechanical apparatus to get around, allowing many other factors to come into play.

 

Even if the data - and the wrapper - are the same, no two files occupy identical positions on a drive: it has been posited that the location matters: slightly different EM signature during drive access maybe? Maybe it's not the circumstance in which the rip took place, but where it lands that counts - in which case ripping the same file twice, in exactly the same way, a week apart might sound different?

 

Beyond that, it's hard to see what else could be going on, even in principle. Which doesn't mean to say we won't all learn something interesting in this regard in the future...

 

[losingmyreligion]

"Bits depend on all kinds of mucky mechanical apparatus to get around, allowing many other factors to come into play."

 

As I posted recently on another thread, a list of possible factors includes:

 

1) "on the fly" FLAC (or other lossless) processing

2) differences in endian-ness (i.e. WAV vs AIFF)

3) meta-data processing (e.g. WAV vs AIFF)

4) differences in drive offset (as it relates to reading from optical drive during initial file copy to disk)

5) files stored on disks in (significant) need of defragmentation

6) files stored on SSDs vs traditional, rotating hard disk

7) files streamed wirelessly from external storage

8) files accessed from NAS (i.e. storage device with it's own operating system)

9) files stored on external drives accessed by USB (which uses computer's processor)

10) files stored on USB thumb drive

 

These are likely only some of the aspects of data files (and how we access them) that could conceivably influence the sound produced by one's DAC, despite that (in most cases) file comparisons would elicit the same checksum, and in all cases the PCM data (i.e. the bits representing music) would be identical.

 

Note: there has been anecdotal evidence reported as to change in sonics attributed to most all of the above factors.

 

clay

 

 

[elcorso]

@khollister & @ jhwalker,

 

There is another technical (o maybe holistic mystery) question:

 

Even if the data (in block mode) feeds my RAM in memory play, and had being always suggested in this CA forum that is no difference between Firewire cables, in my personal case, there is a huge difference when I switched from the stock Firewire 800 cable that came with the external HD, to an inexpensive Verge Labs Firewire cable. It's maybe the block mode data transfered from my external HD to RAM "contaminated" somewhere the RAM data?

 

The RAM bits transfered from the external HD, are not exactly the same bits stored in my HD? Or this bits had added noise and/or jitter, because of the cable?

 

Maybe "bits are bits" is not a valid argument? I don't know.

 

Then, when I changed my internal spinning HD to an SSD, I didn't get any real music reproduction improvement. Even the radiated "noise" from my Mac Mini (with the spinning HD or the SSD), was the same. This measured with the Triefield® Meter Model 100XE. And by the way, both, the Mac Mini and external HD are very noisy, they should be at least 12" (30 cms.) away from any of your other gear and cables, analogue and/or digital.

 

My setup and gear, are listed bellow.

 

 

 

Thanks,

 

Roch

 

[sandyk]

Adding to that, there have now been quite a few documented reports that quite often SQ can be markedly improved by fitting a proper 75 ohm attenuator in line with coax SPDIF output.

Some devices output far too high a level of SPDIF signal.

The claims are that the SQ often further improves to just before the point when lock is lost.There have also been a few

tests done where the recipient did not know the amount of attenuation fitted to the cables that were sent to him for auditioning.

Published wideband C.R.O. photos show that undesirable reflections are often attenuated twice as much as the actual signal.

As Ripley would say, "Believe it, or not !"

Regards

Alex

 

[jhwalker]

I can agree with almost all of the statements above - that it's possible to introduce jitter into the data stream at any point along the way, that electromagnetic fields in the vicinity of equipment can introduce distortion into adjacent equipment, etc. What I will not accept is that two bit-for-bit identical files can sound different consistently when played on the same equipment. Any differences in the impression made during playback must be random - or possibly non-"random", but file source-independent.

 

For example, it is impossible that a file created by ripping thru a megabucks TEAC CD transport can sound different than a file that was ripped on my 10-year-old PC with a buzzy sound card - *if* the two files are proven after ripping to be identical. At that point, they simply *are* identical and they *cannot* produce different results except due to variations in the playback environment. This is a fact that cannot be ignored, even if 100 golden-eared professionals swear an oath otherwise

 

 

[khollister]

Agree 100% with jhwalker. Furthermore there was a post by one of the Metric Halo principles that agreed with my assertion that it is virtually impossible for file corruption to result in the type of sonic differences we are talking about.

 

I am fine with the concept that we do not know how to measure all of the causes of the sonic differences we hear. However I will not consider suspending the laws of physics or mathematics. I can accept the possibility of getting different files from different ripping techniques but not when the files checksum the same.

 

[sandyk]

khollister

But can you accept the possibilty that 2 CDs recorded from the same master , which after ripping have identical checksums, can sound different when played back from the same CD player with error correction working correctly ?

 

SandyK

 

[khollister]

In general, yes. The data is pulled off of a CD in two very different manners in ripping versus Redbook playback. A CD player is accessing the data stream in real time and applying error correction on the fly - very different from what happens in a secure ripper. If the 2 copies of the CD have different errors (media defects, scratches, etc) then it is conceivable that the results when played might be different where a secure ripper would eventually get identical rips from both.

 

However, if you are asking if two defect-free CD's would play back differently, my answer is probably no, unless there was some stochastic element to the playback due to laser tracking or some other item that would trigger the error correction.

 

My problem with this is the inevitable conclusion that if identical digital files (and I do not believe 2 copies of a file that checksum the same can be different from a bit perspective) sound different, then no digital data is reliable and banking, among other things, would not work. Since you claim the differences are repeatable, that rules out a potential stochastic/random playback effect. If the bits are the same, how do you possibly explain how a specific DAC reconstructs a different analog waveform from the same bit stream?

 

The only possible theory that could support this is a previously undiscovered "hole" in the principle of a checksum, i.e. checksums are not unique. I suppose it could be possible that audiophiles have detected a flaw in the mathematics of this, but ...

 

Even if this were the case, I'm still in denial over how errors (induced by noise) in the ripping process produce the type of changes in a digital file that people are describing. They can only be the result of a correlated, systemic alterations of low order bits in every single word in the audio file. That has to be deterministic, not random, and that strikes me as even more incredulous than discovering a flaw in the theory of checksums.

 

Sorry, but I'm an Electrical Engineer, work in the software industry these days, a musician and an audiophile for over 30 years (and, yes, I think cables matter, there is some magic in tubes we don't know how to measure, etc) - I guess I know just enough and have too much experience to believe in the tooth fairy (sorry - that's probably a cheap shot).

 

I absolutely believe that there are audible effects that we do not know how to quantify or predict, but I can't accept things that defy fundamental physical and mathematical principles.

 

[1audio]

I also agree that if the data is the same the playback will be the same. However real time issues can affect the sound. The biggest offenders are the noise links from box to box and the power supplies.

There are two primary modes of noise coupling, normal mode (noise between conductors on the cables) and common mode (noise conducted through the cables and associated power supplies and external power (AC etc.) connections) and a third more difficult mode, RF through the air.

 

I did some measurements of drive noise and was surprised to see that an SSD can have higher noise than a hard disk. It was an isolated test with a mimimal sample size it should not be extended to a generalization. However the noise was not inconsiderable.

 

I have also looked at the common mode noise issue and there are many paths for noise to circulate through a computer audio system and "pollute" the analog chain. Before the proliferation of computers and personal digital electronics it was obvious when a noisy digital device was in the listening room. I remember shutting off the display on a JVC CD player and hearing a substantial improvement in the sound (even vinyl). I have replaced digital clock generators in super high end vinyl transports (Turntables) with analog oscillators with substantial sonic benefits. On my first DAC (the ENTEC Number Cruncher) I provided a method of locking it in reset that shut down the internal clocks, to avoid degradation in any analog playback when its not in use. In the above cases its radiated noise that was the culprit, no electrical connection to the circuits existed.

 

Before attribution of sonic differences to unseen magic hidden in the bits, invisible to the data itself, these well understood effects need to be resolved and controlled. Reducing them will make a major improvement regardless, anyway. It will also prove to be very difficult to remove some of the noise sources or noise paths. For example, "galvanic isolation" doesn't mean that there is no capacitive common mode coupling. With high data rates and their associated high frequencies it takes very little capacitance to couple a lot of energy between two boxes. The network cable, a USB drive with an external supply (analog or digital) both represent multiple pathways for noise. The drive supply connection to AC and from AC through the PC supply to the other end of the USB cable is a circuit for noise. The same for the network cable and its other end (a router for example). Going wireless just moves the noise from conducted to radiated with a different spectrum if modulation to mess the audio up. (This all sounds like an advertisement for Ayre's heavy aluminum construction, which isn't intentional, but that does work well).

 

 

[sandyk]

"However, if you are asking if two defect-free CD's would play back differently, my answer is probably no..."

 

http://www.cdjapan.co.jp/grapes/-/archives/2008_11_14__blue_spec_cd

Also, check out SHM CD.

 

I have a comparison copy of the CD at the link below , as well as a couple of other comparison Sony BluSpec CD sets.

I can hear clear differences between both CDs in the set using an internal DVD writer, and an internal LG BR writer, as well as 2 different SACD/DVD-A players.

The differences can still be heard when played with an upmarket player such as the AU$4,500 (originally) Marantz SA11, but are not as obvious.

This is exactly as Sony intended, and these comparison sets of the same material from the same masters, were used to introduce the format in Japan by Sony. Both comparison CDs in the sets have identical check sums when ripped using EAC, and I can still hear differences between them AFTER they are ripped, although not as obvious compared to normal playback.

I am NOT into imagining these differences either, as other people on the other side of the world with better than average gear have been able to hear the differences after they downloaded them, even though they had no knowledge of what to expect.

There are now a large number of BluSpec CD releases.

SandyK

 

http://www.cdjapan.co.jp/detailview.html?KEY=SICP-20048

 

[Audio_ELF]

SandyK, I don't think comparing a Blue Spec CD to a standard CD is quite what khollister meant. Two copies of the same CD *should* sound the same as long as both are pristine (ie. no error correction / interpolation needed).

 

Equally two files with identical checksums should not be able to sound different on playback (if both are played back from the same drive / memory stick) regardless how each was ripped from the CD.

 

Eloise

 

[dsnyder]

Hello,

 

Never heard of blu-spec cds before. Cool. Thanks for sharing the link! I have noticed that blu-ray discs are a bit harder to scratch than standard CD or DVD discs, and perhaps using the same manufacturing process as blu-ray can help bring greater economies of scale. This format could even become the next "XRCD" since it sounds like the real-time playback performance improvements are similar.

 

I don't doubt that blu-spec cds sound a little better than standard issue CDs created from the same PCM master when played in an audio CD player--even a very nice one. But just like XRCD rips, the PCM data will be bit-for-bit identical to a standard pressing made from the same master (ignoring a hundredth of a second or so in difference between playback start times, assuming non-defective discs in good physical condition and a properly configured EAC/dbPoweramp workstation). One of the things that I love about computer audio is that I don't have to care so much about special pressings of physical CDs or fiddling with products like the Accoustic Revive RD-3 disc demagnetizer to be sure that I'm getting most of the bits that I paid for. No doubt that stuff like this makes a difference in a physical audio CD player, but once you have reliable ripping process, you're free to put these up on ebay. :-D

 

I would not accuse you of imagining the differences that you have heard but I respectfully suggest that the explanation for these differences is something other than what you have described in this and other posts. Without standing over your shoulder and watching your process for CD rip and playback, I can only imagine...

 

-- David

 

 

[sandyk]

Eloise

I am making the point that there is much more to the whole subject than most people, especially hard nosed Objectivists wish to believe.I expect that much of the knowledge amassed

by companies such as Sony, who are the co-developers of the CD format, is kept under wraps for commercial reasons.

They obviously knew what they were doing when they exploited the advantages of the BluSpec format.

I disagreed with you almost 2 years ago in this forum , and continue to disagree with you on this matter.

In the meantime, there have been a lot more posts on this subject in various forums worldwide.Even in CA, there are now a lot more people also speaking out, which is why I decided to rejoin this debate.This was also in part due to a recent question in another CA thread, where I was quoted, despite the original thread now being quite old.Many people these days use one of the most revealing test instruments available. Their EARS !

Even 3 highly respected Recording Engineers are saying the same in another thread here, about audible degradation on playback of .flac files compared with uncompressed formats.

That was also one of my original contentions back then.

I suspect that you were one of those who refused to believe that back then too ?

Many people worldwide are now reporting hearing things that people like yourself consider impossible.

Regards

Alex

 

[Guest Claude]

How long ago was it Alex? Havent heard from you in a long time. Good to see your still at the digital forefront.

I remember how Peter had some very reasonable (while not always logical and always controversial) ideas on what might happen. I do not remember such a wonderful summary of possible reasons for this phenomenon (well done Clay (you are not the "Clay" Clay or did you change your name?)). And I dont remember the excellent comments on noise like the one from Damian. I always thought that galvanic isolation would be a complete isolation (just as the word implies or does it mean something different in English?). Very good explanations.

Yes, I hear it too: wav sounds different from aiff from flac and yes different rips with the same checksum sound different. I did not go into the other experiments, because this somehow got frustrating to see how many things influence sound.

I also remember how so many people insisted, that first it has to be proven, that there can be an existing difference before they are willing to listen to the files supplied by Alex (SandyK).

Alex if you still supply this stuff which are nightmares made off - let them go! Since I last compared some files my system has improved and I think I can take it now. ;-)

And if nobody hears a difference. Wonderful! I wish I had never heard a difference. This would keep me from wondering about how my complete library would sound if I converted it to wav (which I refuse to do because of convenience).

 

PETER ST - are you there? Have you come up with explanations others then the ones stated here? I am sure you have! Let us participate.

 

All the best from Berlin

Claudius